From: Vermaas, Josh (vermaasj_at_msu.edu)
Date: Tue Nov 30 2021 - 11:10:03 CST
Hi Vlad,
In addition to the great points Axel and Giacomo have made, I’d like to point out that the 8160 is a 24 core processor, and that there are likely 2 of them on a given node. In these two socket configurations, where there are two physical CPU dies, I’ve often found that the best performance is achieved when you treat each socket as its own node, and allocate 2x the number of “tasks” as you have nodes. That way, each SMP task gets placed on its own socket. If you don’t, each node is trying to get all 48 cores across both sockets to work together, which ends up saturating the UPI links between the sockets, and can be detrimental to performance.
This is usually a bigger problem for SMP-based builds. In my experience, CPU-only systems benefit from MPI based builds, where the number of tasks is equal to the number of CPUs. Usually this is a performance win for modestly sized systems at the expense of scalability for really big systems.
-Josh
From: <owner-namd-l_at_ks.uiuc.edu> on behalf of Vlad Cojocaru <vlad.cojocaru_at_mpi-muenster.mpg.de>
Organization: MPI Muenster
Reply-To: "namd-l_at_ks.uiuc.edu" <namd-l_at_ks.uiuc.edu>, Vlad Cojocaru <vlad.cojocaru_at_mpi-muenster.mpg.de>
Date: Tuesday, November 30, 2021 at 10:18 AM
To: Giacomo Fiorin <giacomo.fiorin_at_gmail.com>, NAMD list <namd-l_at_ks.uiuc.edu>, Axel Kohlmeyer <akohlmey_at_gmail.com>
Cc: HORIA-LEONARD BANCIU <horia.banciu_at_ubbcluj.ro>
Subject: Re: namd-l: NAMD performance on a supercomputer with Intel Xeon Platinum 8160 and 100Gb Intel Omni-Path Full-Fat Tree
Thanks for your thoughts !
One thing that seemed weird during our tests on this site was that the performance and parallel scaling rapidly degraded when using all 48 cores available per node (2 CPUs with 24 cores each) . We actually saw negative scaling after as few as 16 nodes. Then, when using 47, 32, and 24 cores/node, we got better parallel efficiency to higher node count, with the best efficiency obtained using just half of the core available on each node (24). At the end, when running on 48 nodes, we achieved the most ns/day when using 24 cores/node. However, the resources requested in the project we needed to calculate using all 48 cores/node regardless of how many we are actually using.
I haven't experienced anything like this on other sites (similar systems, same configuration files). Normally using all cores available per node has always given the best performance. So, I am wondering whether there is anything obvious that could explain such a behavior ?
Best
Vlad
On 11/30/21 15:33, Giacomo Fiorin wrote:
Something in addition to what Axel says (all of which is absolutely true, even the counter-intuitive part about making the single-node performance artificially slower to get through the bottom-most tier of technical review).
One possible issue to look at is how the cluster's network is utilized by other users/applications. In a local cluster that I use, the InfiniBand network is also used by the nodes to access data storage and there are many other users processing MRI, cryo-EM or bioinformatics data (all embarrassingly-parallel by design). So the InfiniBand network is constantly busy and does not necessarily offer very low latency for NAMD or other message-passing applications.
Something that helped in that case was building Charm++ on top of the UCX library instead of IBverbs directly. I am wholly unfamiliar with the details of how UCX works, but in essence it provides better utilization of the network when the ratio of compute cores vs. network links is high. If the cluster's staff has a copy of UCX, try that. It wasn't easy to build, but it paid off specifically for those runs that were communication-bound.
The main significant addition in 2.15 is the AVX-512 tiles algorithm, which would help with the most expensive Intel CPUs like those, but would also make the computation part faster with the caveat that Axel mentioned.
Giacomo
On Tue, Nov 30, 2021 at 6:16 AM Axel Kohlmeyer <akohlmey_at_gmail.com<mailto:akohlmey_at_gmail.com>> wrote:
Actually, if you optimize how NAMD is compiled better than the system provided executable, your parallel efficiency will go down. Please recall Amdahl's law: the parallel efficiency is determined by the relation of time spent on parallel execution and serial execution.
A better optimized executable will spend even less time computing and thus have more parallel overhead.
To get better parallel efficiency, you have to avoid or reduce all non parallel operations like output or use of features like Tcl scripting or make your computations more expensive by increasing the cutoff or the system size or make the executable slower by compiling a less optimized version.